UvA-DARE (Digital Academic Repository) The role of analogy in language change: Supporting constructions De Smet, H.; Fischer, O. DOI 10.1017/9781316091746.011 Publication date 2017 Document Version Submitted manuscript Published in The Changing English Language Link to publication Citation for published version (APA): De Smet, H., & Fischer, O. (2017). The role of analogy in language change: Supporting constructions. In M. Hundt, S. Mollin, & S. E. Pfenninger (Eds.), The Changing English Language: Psycholinguistic Perspectives (pp. 240-268). (Studies in English Language). Cambridge University Press. https://doi.org/10.1017/9781316091746.011 General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons). Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible. UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl) Download date:29 Sep 2021 The role of analogy in language change Supporting constructions Hendrik De Smet and Olga Fischer (KU Leuven / University of Amsterdam/ACLC) 1. Introduction It is clear from the literature on language variation and change that analogy1 is difficult to cap- ture in fixed rules, laws or principles, even though attempts have been made by, among others, Kuryłowicz (1949) and Mańczak (1958). For this reason, analogy has not been prominent as an explanatory factor in generative diachronic studies. More recently, with the rise of usage-based (e.g. Tomasello 2003, Bybee 2010), construction-grammar [CxG] (e.g. Croft 2001, Goldberg 1995, 2006), and probabilistic linguistic approaches (e.g. Bod et al. 2003, Bod 2009), the interest in the role of analogy in linguistic change -- already quite strongly present in linguistic studies in the nineteenth century (e.g. Paul 1886) -- has been revived. Other studies in the areas of lan- guage acquisition (e.g. Slobin, ed. 1985, Abbot-Smith & Behrens 2006, Behrens 2009 etc.), cog- nitive science (e.g. work by Holyoak and Thagard 1995, Gentner et al. 2001, Gentner 2003, 2010, Hofstadter 1995), semiotics (studies on iconicity, e.g. Nöth 2001), and language evolution (e.g. Deacon 1997, 2003) have also pointed to analogical reasoning as a deep-seated cognitive principle at the heart of grammatical organization. In addition, the availability of increasingly larger diachronic corpora has enabled us to learn more about patterns’ distributions and frequen- cies, both of which are essential in understanding analogical transfer. Despite renewed interest, however, the elusiveness of analogy remains. For this reason, presumably, Traugott (2011) distinguishes between analogy and analogization, intended to sepa- rate analogy as (an important) ‘motivation’ from analogy as (a haphazard) ‘mechanism’. The 1 Under analogy we understand both the recognition of simple item similarity as well as relational (structur- al) analogy, see Behrens, but pace Felser??, this volume. distinction is deemed necessary since “much analogical thinking never results in change” (ibid. p.25). Analogization thus focuses on the results of analogical thinking. Another attempt to mitigate the ‘unruliness’ of analogy is found in Bybee and Beckner (2014), who consider categorization more important than analogy. This goes against most psy- cholinguists’ conviction that categorization itself can only be the result of analogical thinking (see e.g. Chalmers et al. 1992, Gentner 2010, Gentner et al. 2011). Bybee and Becker note, nev- ertheless, that “the much rarer change of analogical extension” is “not accomplished by analogi- cal reasoning but rather by changes in existing categories” (p.506). Examining “a wide range of changes that have been called ‘analogical’”, they find that “all [changes] fall under the umbrella of changes in categorisation and that a separate mechanism of change dubbed ‘analogical ’is not necessary”. They illustrate this with the case of the past tenses of strike and dig, for which “no proportional model […] is possible” because these two verbs do not share the phonological characteristics that are deemed necessary to make the analogy. That is, they do not share “the property of ending in a velar [nasal] consonant” of the wider cling, swing class (with past tense - ung), to which new members were added (sling-slung, hang-hung) (ibid.). It seems to us that analogy is interpreted too narrowly here. Analogy not only concerns a similarity between concrete (in this case phonological) forms, and a similarity between abstract patterns (in this case verbs and their past tenses) but also local similarities in function or meaning (cf. the idea of “structural mapping” [Gentner and Namy 2006, Gentner and Smith 2012]). As Anttila (2003), Itkonen (2005) and others have made clear, and as CxG linguists emphasize, analogy always involves a combination of form and meaning. The new past tenses of strike and dig can be explained by analogy when one looks more closely at the full synchronic circum- stances operating at the time of their first occurrence. When we look at it in terms of analogy, the new past could well have been supported semantically by the past tenses of cut and stick ‘stab’ (i.e. cut and stuck), which like strike and dig convey a cutting or stabbing movement. As to form, they share the short vowel and the plosives. Like struck, stuck and dug are attested in the OED from the sixteenth century onwards. In addition, it must be mentioned that it is quite likely that struck was an independent phonetic development (cf. Ekwall 1965, §§81, 215, Hogg 1988): the ME form was [strɔ:k] (from OE [ɑ:]) but with a variant [stro:k], which with other ME [o:]- sounds developed to [u:] in the Great Vowel Shift and was later shortened to [u] > [ʌ] before plosives. If this is correct, the form struck itself could have strenghened the analogy. And final- ly, a more general iconic principle may also have played a role, that of the ‘principle of quantity’. It is noteworthy that cut, struck, dug and stuck all convey a highly telic, brief movement, which is more appropriately conveyed by a phonologically shorter form (cf. also the ‘ideophonic’ ar- gument put forward in Hogg [1988], where he considers the past tense snuck in relation to dug and struck). This development shows that one need not always have a very frequent pattern for analogical extension; a local pattern, if strengthened semantically, may do the job too.2 Another example of interaction between multiple local analogies – this time affecting syntactic structure – can be observed in Dutch and involves the extension of a reflexive pronoun in verbs denoting psychological activities. The category contains verbs such as zich[REFL] herin- neren ‘to remember’, zich[REFL] realiseren ‘to realise’, zich[REFL] ergeren ‘to be annoyed’, and has been joined quite recently by verbs such as beseffen ‘to realise’, irriteren ‘to be irritated’, which are now also used reflexively by younger speakers. The analogy here concerns not only the semantic similarity but also the fact that some of these verbs share a causative structure. Thus, there is both causative dit herinnert me[OBJ] eraan dat… / dit ergert me[OBJ] ‘this reminds me that…/ this annoys me’, and reflexive ik herinner me[REFL] dat…/ ik erger me[REFL] eraan dat… ‘I remember that … / I am annoyed that’, resulting in the causative verb irriteren (dit ir- riteert me[OBJ] ‘this irritates me) to also develop a reflexive construction: ik irriteer me[REFL] ‘I am irritated’. Thus a network of analogies, involving both causative and reflexive verbs express- ing mental activities, may lead to local change. What these examples show is that if analogy is to be properly understood, its operation must be seen against the background of complex constructional networks capturing the myriad relations between constructions. In this respect, Abbot-Smith & Behrens (2006) propose the no- tion of “supporting constructions” in language acquisition, to explain how it is that, for instance, German children learn certain constructions earlier than others. In Abbot-Smith and Behrens (2006: 1019), it was found that a ‘supported construction’ was acquired earlier and faster than a non-supported one if the source and target constructions “share[d] lexical or morphological sub- parts”. They show that the sein-passive (i.e. the perfect passive formed with HAVE-been+past par- ticiple in English) is acquired earlier than the werden-passive (the non-perfect passive formed with BE+past participle) due to the fact that a lexical-morphological and highly frequent subpart 2 Barðdal (2008) makes clear that analogy may occur even with low type frequency provided there is strong semantic coherence. of the construction is already familiar to German children in the form of the perfect construction with sein+past participle. Interestingly, they also show that the acquisition of a target construc- tion can be hindered if two constructions “share an identical semantic-pragmatic function”. We believe that the notion of ‘supporting construction’ may also help to explain how constructions spread analogically in diachrony. As we will show, the ‘construction conspiracy hypothesis’ − the term Abbot-Smith and Behrens use − can be extended to language change. It has been observed, for instance, that an analogical extension is the more likely, the more its out- come resembles one or more already existent patterns (Bybee & Slobin 1981; Fischer 2011, 2015; De Smet 2012, 2013).